Implications of the directivity of railway noise sources for their quantification using conventional beamforming

This paper investigates the effect of the directivity of railway noise sources on the results of an identification procedure based on beamforming using a microphone array. Usually when performing pass-by noise tests, a single-microphone noise spectrum is obtained for a time window corresponding to the length of the whole train, or of a single vehicle. In this context, a source quantification algorithm should be able to evaluate the contribution of each noise source over this time window. However, different railway noise sources have different directivities, and it is shown that these need to be taken into account to achieve accurate source quantification. By making use of monopoles, dipoles and quadrupoles, it is shown that a different compensation is needed according to the directivity. For the particular case of the noise radiated by the rail, this has a complex directivity pattern that is only partially captured by a microphone array. It is demonstrated that the overestimation of the wheel contribution found in previous research may be attributed to a misinterpretation of part of the rail contribution from the beamforming map.